Justice Kassio Nunes Marques, president of Brazil's Superior Electoral Court (TSE), the body that oversees elections in the country, has scheduled for next Tuesday (September 1) the trial of a case brought by the Workers' Party (PT) against a video of former president Jair Bolsonaro produced with artificial intelligence. The material was shown at the convention that confirmed Senator Flávio Bolsonaro, the former president's son, as the PL party's presidential candidate. According to G1, justices indicate the ruling should set the court's interpretation of AI-generated content in the 2026 elections.
The TSE resolution published this year bans the use of the technology "to harm or favor a candidacy, of synthetic content in audio, video or a combination of both," even with the authorization of the person portrayed. According to G1, the central question is whether the material can mislead voters into believing that a given fact occurred, with an appearance of authenticity. That line would be classified as a deepfake and would be banned, whether the content is positive or negative.
A majority of the justices have voted along these lines in cases decided in the court's virtual plenary. TSE vice-president André Mendonça proposed a thesis with this interpretation when he ordered the removal of a video linking Flávio Bolsonaro to former banker Daniel Vorcaro, who was jailed in Operation Compliance Zero. Justices heard on condition of anonymity by G1 signal that the video shown at the convention could be ruled irregular and that any punishment would likely be a fine. Bolsonaro is under house arrest serving a 27-year-and-three-month sentence in the so-called coup plot case and is barred by the courts from issuing political-electoral statements, including through third parties.
The PT's case and the defense
In its filing, the PT argues that, although the video states it is a simulation, TSE rules ban the use of synthetic content to favor a candidacy, even when there is authorization to reproduce a person's image or voice.
"It is, therefore, a device that amplifies early electoral propaganda and, at the same time, breaches the ban on the use of artificial intelligence aimed at creating synthetic content to replace the image and voice of a living person, in addition to circumventing the judicial restriction imposed on the former president through the use of synthetic reproduction technology."
Flávio Bolsonaro's campaign told the TSE that the video did not violate electoral law because it did not hide the use of the technology. In the footage, the avatar identifies itself as a "simulation" of the former president's image and voice. The defense compared the resource to the masks and paper puppets commonly used in campaigns. "There was only the creation of a technological puppet, all with the proper labels and warnings, without any falsification or inducement into error," the lawyers wrote, asking the TSE not to ban AI from the race entirely.
The PT rebutted that thesis in a new petition, CNN Brasil reported. The party cited the precedent of Evandro Leitão (PT-CE), fined R$ 15,000 (about US$ 3,000) by the TSE itself after publishing AI videos featuring Barack Obama and Taylor Swift during the 2024 municipal elections. The party also pointed to legislation in the European Union, South Korea and the US states of Texas and Minnesota, where labeling a video as a deepfake does not authorize its use. "The very normalized circulation of realistic simulations of public figures erodes the probative value of the audiovisual record," the PT lawyers argued.
TSE justices already met in August to seek common ground on the limits of AI before the merits trial, according to CNN Brasil. The vote is set for Tuesday (September 1) in a plenary session chaired by Kassio Nunes Marques, and should indicate whether the disclosed use of AI remains banned or whether the court will revise its guidelines.